Risk-based Design of Experiments
Also known as: Risk-based DoE, risk-informed experimental design, risk-prioritized DoE, RB-DoE
Risk-based design of experiments (RB-DoE) integrates formal risk assessment — typically using tools such as FMEA or fault tree analysis — with classical experimental design to prioritize which process or product factors are most critical to investigate. Rather than treating all candidate factors equally, this approach ranks factors by their risk priority number or likelihood of affecting quality, safety, or reliability, then allocates experimental runs preferentially to high-risk factors. It is widely used in pharmaceutical development, chemical process engineering, and manufacturing quality management.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use RB-DoE when you face many potential experimental factors but limited run budget, and some factors carry substantially higher consequences if not controlled (safety, regulatory, or product-quality implications). It is especially appropriate in pharmaceutical process development under ICH Q8/Q9/Q10 frameworks, medical device manufacturing, chemical process scale-up, and any regulated industry where a risk-based approach to process understanding is required. Do not use it as a substitute for classical DoE when risk information is absent or all factors are equally critical — in such cases a standard factorial or RSM approach is simpler and equally valid. Avoid it when the risk scoring itself is highly uncertain or politically driven, as a flawed risk ranking will produce a biased experimental design.
Strengths & limitations
- Focuses experimental resources on the factors most likely to affect safety, quality, or regulatory compliance.
- Produces a defensible, documented rationale for design choices — essential in regulated industries (pharma, medical devices, aerospace).
- Integrates well with existing quality-management tools (FMEA, fishbone diagrams, cause-and-effect matrices).
- Reduces the risk of missing a critical interaction by ensuring high-risk factor pairs are not aliased in the design.
- Provides a structured feedback loop: experimental results update the risk model, improving future risk assessments.
- The quality of the experimental design is directly dependent on the quality of the prior risk assessment; a poor FMEA produces a misallocated design.
- Risk priority scoring (especially RPN in FMEA) is often subjective and may not reflect true probabilities or consequences.
- Additional complexity in design planning may require specialized expertise in both risk management and statistical experimental design.
- Not well suited to purely exploratory experiments where no prior risk information is available.
Frequently asked
How is RB-DoE different from a standard design of experiments?
Standard DoE treats all candidate factors symmetrically during design selection, choosing designs based on statistical efficiency criteria alone. RB-DoE adds an upstream risk-ranking step that determines which factors are included, at what resolution, and with what degree of replication, based on their assessed risk priority. The experimental execution and analysis steps are then identical to classical DoE; the difference is entirely in the factor-selection and design-choice logic.
Which risk assessment tool works best with RB-DoE?
FMEA (Failure Mode and Effects Analysis) is the most commonly used tool, particularly in pharmaceutical and manufacturing settings, because it yields a numerical RPN that can directly rank factors. Cause-and-effect (Ishikawa) matrices are also popular for initial screening. For safety-critical applications, fault tree analysis or bow-tie analysis may be preferable because they model failure propagation pathways more rigorously. The choice depends on the industry context and the depth of prior process knowledge.
Can I use RB-DoE in an early-stage exploratory study where little is known about the process?
RB-DoE is least useful when prior process knowledge is very limited, because the risk assessment that drives factor prioritization requires at least a qualitative understanding of potential failure modes. In early exploration, a standard screening design (Plackett-Burman, 2^k-p fractional factorial) is more appropriate. RB-DoE becomes valuable once a preliminary understanding of failure modes allows a defensible risk ranking of factors.
Is RB-DoE required by regulatory agencies?
In pharmaceutical development, ICH Q8 encourages but does not mandate a specific experimental design method; it does require that the development approach be science- and risk-based. Many regulatory submissions use RB-DoE as evidence of a systematic, risk-informed process understanding. In other regulated industries (medical devices, aerospace) analogous quality-system standards encourage risk-based process validation, which naturally leads to RB-DoE in practice.
How do I handle a factor that has a high risk score but is difficult to vary experimentally?
If a high-risk factor is practically difficult to vary (e.g., raw material supplier, environmental conditions), it should be included as a blocking factor or a noise factor in the design rather than a controlled design factor. This preserves awareness of its risk while acknowledging the experimental constraint. Alternatively, a separate smaller confirmatory study focused solely on that factor may be conducted alongside the main design.
Sources
- Myers, R. H., Montgomery, D. C., & Anderson-Cook, C. M. (2016). Response Surface Methodology: Process and Product Optimization Using Designed Experiments (4th ed.). Wiley. ISBN: 978-1118916018
- International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH). (2009). Pharmaceutical Development Q8(R2). ICH Expert Working Group. link ↗
How to cite this page
ScholarGate. (2026, June 3). Risk-based Design of Experiments. ScholarGate. https://scholargate.app/en/experimental-design/risk-based-design-of-experiments
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Design of experimentsExperimental design↔ compare
- Failure Mode and Effects AnalysisExperimental design↔ compare
- Fault Tree AnalysisReliability↔ compare
- Response Surface MethodologyExperimental design↔ compare